**** BEGIN LOGGING AT Sun Aug 05 02:59:59 2018 Aug 05 08:09:08 apparently not even the google pixel 2 supports 802.11w Aug 05 08:09:19 let alone my nexus 6p Aug 05 08:17:04 i guess google just doesn't give a damn about 802.11w Aug 05 08:22:07 huaracheguarache: is there a way to show if a client is using 802.11w with e.g. wpa_cli? Aug 05 08:22:27 i have a 6p connected to an ath9k AP w/802.11w "enabled" Aug 05 08:22:37 * m4t is curious now Aug 05 08:22:53 maybe tcpdump w/monitor mode will show.. Aug 05 08:25:29 my test simply was to make 802.11w mandatory and to test which vclients were still able to connect Aug 05 08:28:05 oh yeah. there's more than just a bool Aug 05 08:30:08 * m4t set it to 2 "required". nexus reconnected :/ Aug 05 08:30:27 i did the same thing as pkgadd, but i only tested the ac network Aug 05 08:30:59 the ipad connects without any issues Aug 05 08:31:43 oh, mine is 802.11an :/ Aug 05 08:31:55 i dont have any of that fancy new ac kit Aug 05 08:33:01 802.11w is actually a mandatory part of the ac certification, but the nexus 6p probably isn't certified Aug 05 08:34:10 not just you https://www.reddit.com/r/Nexus6P/comments/52hs8v/problem_with_protected_management_frames/ Aug 05 08:34:25 uhm, i am not running stock though. idk if that'd make a difference, i think they're using stock vendor partition. Aug 05 08:34:40 are you running copperheados? Aug 05 08:34:45 nope, lineageos Aug 05 08:34:48 ok Aug 05 08:35:54 yeah, i just checked and the nexus 6p isn't certified by the wifi alliance Aug 05 08:38:50 2.4ghz doesn't work for me =( Aug 05 08:39:15 so it might have something to do with lineageos Aug 05 08:44:09 maybe it's buggy and google disabled it or something Aug 05 08:47:34 that might be. pkgadd: when 802.11w is set to optional do clients that support it automatically connect with it enabled? Aug 05 08:51:09 Hi, is it possible that the snapshot imagesbuilders are sometimes out of sync with the packages, resulting in trouble when installing kmods which are fetched from remote? Aug 05 08:56:01 Here is a buildlog showing the problem https://as-test.stephen304.com/download/faillogs/faillog-4ded89cc3198.txt Aug 05 09:03:04 morning guys Aug 05 09:03:22 i am running the last LEDE on my tp-link wdr4300 Aug 05 09:03:39 jow: you got an idea? Aug 05 09:03:41 tl-wdr4300-v1-squashfs-sysupgrade.bin Aug 05 09:03:56 this is the image i want to update to the new openwrt release, right? Aug 05 09:04:06 r0bert: meaning 17.01.5 or 18.06.0? Aug 05 09:06:20 i am on lede 17.01.4 and want to upgrade to 18.06 Aug 05 09:06:41 https://downloads.openwrt.org/releases/18.06.0/targets/ar71xx/generic/openwrt-18.06.0-ar71xx-generic-tl-wdr4300-v1-squashfs-sysupgrade.bin Aug 05 09:06:47 this is, what i want, right? Aug 05 09:09:42 r0bert: yes, i believe so Aug 05 09:12:40 m4t: nope, still doesn't make much sense. could it be that your local config is messed? removing the openwrt tmp directory might work as well Aug 05 09:12:50 blogic: 1 of 2 APs pooped itself again after less than 36h uptime Aug 05 09:13:20 mkresin: hmm i did try the commit on a distclean tree before making the PR Aug 05 09:14:51 config is very simple too. select ath79, select TARGET_ALL_PROFILES, select CONFIG_TARGET_PER_DEVICE_ROOTFS Aug 05 09:15:23 m4t: can we switch to query? Aug 05 09:15:28 sure Aug 05 09:15:29 everything went fine \o/ Aug 05 09:15:43 is openwrt 18.06 also codenamed with a cocktail? Aug 05 09:16:18 I think we voted against having release names Aug 05 09:17:36 that's too bad Aug 05 09:18:47 (i am using openwrt since white russian and i think it made it less more technical and a little bit more fun :P) Aug 05 09:19:45 vote against it was unanimous Aug 05 09:21:01 would it be better to start the config from scratch when upgrading to 18.06? Or can one keep the 'keep settings' checkbox set? Aug 05 09:21:21 stintel: do you have 802.11w set to forced or optional? Aug 05 09:21:42 optional, as I have some clients that don't support it Aug 05 09:22:22 bookworm: there are people who advise to start from scratch, for me that would be a major pita Aug 05 09:22:37 yeah, that's why I ask... I'd rather avoid it Aug 05 09:22:42 i'll have to wait and see if my network will crap itself eventually Aug 05 09:23:04 I'd suggest to keep settings, and if there is unfixable fallout, you can restore defaults and start from scratch Aug 05 09:23:05 but if the answer would have been, god no, that'd be stupid I would have done so ;) Aug 05 09:23:24 thanks for the advice Aug 05 09:23:30 but if it "just works", .. :) Aug 05 09:28:01 morning Aug 05 09:30:13 wigyori: o/ Aug 05 09:31:07 \o Aug 05 09:32:14 aparcar[m]: these packages are all built by phase1 builders: they should always be in sync Aug 05 09:34:42 f00b4r0: thank you, I'll do some more debugging Aug 05 09:34:49 aparcar[m]: actually scratch that, some of your failed deps are from packages that aren't built by phase1 Aug 05 09:35:02 ppp for instance Aug 05 09:35:33 aparcar[m]: note that each "released" kmod version is archived in e.g. https://downloads.openwrt.org/snapshots/targets/mvebu/cortexa9/kmods/ Aug 05 09:35:47 precisely to work around that situation, AIUI Aug 05 09:38:15 aparcar[m]: if I'm not mistaken, anything that's under snapshots/targets/// is built by phase1 builders and thus *should* always be in sync Aug 05 09:39:08 f00b4r0: thank you very much for the details. I'll check if my script is broken and thus the snapsho imagebuilders are much to old Aug 05 09:39:19 bu the content of snaphots/packages/ is built by a different set of builders and I can see a situation where they'd be out of sync with the former Aug 05 09:40:12 jow would know more about the details, I'm only familiar with how phase1 builders operate. What happens on the public-facing server is black magic to me :) Aug 05 09:41:54 aparcar[m]: in any case I think the 'kmods/' folder is meant to address exactly this problem Aug 05 09:51:51 I'll check what went wrong, currently most of my image server builds fail due to the kmod problem Aug 05 10:27:19 karlp: fixed the time::hires dependency Aug 05 10:27:47 stintel: that's a little revisionist to say voting against it was unanimous Aug 05 10:28:57 jow: nice. thanks. Aug 05 10:29:48 still love to see this build monitor tool that was porposed when this time thing came in :) who was it? ettienne? Aug 05 10:30:55 would've liked if that that was implemented to fail gracefully Aug 05 10:31:05 this thing burned several hours of resources for several developers Aug 05 10:31:24 yeah, I've seen all the pain of time, gtime, my system doesn't have the right option to time. blah. Aug 05 10:31:24 which is a little bit too much for "hey, here's a little noninvasive change to print time" Aug 05 10:32:04 also the format sucks, it produces stderr output when there's no err etc. Aug 05 10:33:51 okay enough ranting, its not as if it was written with bad intentions in mind Aug 05 10:37:48 aparcar[m]: from your log it seems that you cache the imagebuilder Aug 05 10:38:01 that will fail if the kmods are updated since then Aug 05 11:09:43 jow: jep, I think this meta imagebuilder script failed to detect new IB versions Aug 05 11:27:14 jow: what do you think in general of a "meta imagebuilder" which simplies the handling of multiple IBs? Aug 05 11:41:30 * ldir uninstalled gnu-time Aug 05 11:53:46 does ath10k support 802.11w in client mode? Aug 05 14:39:09 huaracheguarache, -ct should Aug 05 14:41:51 greearb: maybe i'll give it a spin since the regular ath10k doesn't seem to support it Aug 05 14:48:47 huaracheguarache, you might need the -htt-mgt version, should be available in latest openwrt I think Aug 05 14:49:57 stintel: ping Aug 05 14:59:03 greearb: to compile ct i need to select it in menuconfig under 'firmware' and 'kernel modules -> wireless drivers', correct? Aug 05 14:59:26 Hi! Aug 05 15:00:12 A dumb opkg question if I may: Is there a way to have opkg verify that the files installed by a package have not been modified after installation? Aug 05 15:01:52 I manually patched one file, but I can't remember which. Aug 05 15:05:20 opkg doesn't check integrity afaik. Aug 05 15:06:34 carldani: uhm, check mtime on the file(s) Aug 05 15:06:46 (with find and/or ls) Aug 05 15:11:22 greearb: brb, think i figured it out. Aug 05 15:11:58 m4t: I feared that it wouldn't be easy, but I'll try that. Thanks. Aug 05 15:17:50 carldani: back up the files, remove the package,reinstall it, diff Aug 05 16:11:46 greearb: i didn't manage to make it work. when i set 802.11w to required the client doesn't connect Aug 05 17:57:17 huaracheguarache: on the clients I have to set pmf=1 and ieee80211w=2 in wpa_supplicant.conf Aug 05 18:02:37 pkgadd: is that with qca ath10k or ct? Aug 05 18:13:56 hold on, i think i dc'ed Aug 05 18:15:46 pkgadd: is there a wpa_supplicant.conf file in openwrt? Aug 05 18:16:41 huaracheguarache: you asked me about client-side changes, that's what I need to set on my client (ath9k, ath5k, etc. don't have any client with 10k) Aug 05 18:18:37 ah, right. sorry, got a bit sidetracked since i tried to make it work in client mode on my archer c7 Aug 05 18:56:12 wifi changes up comming in k 4.19 Aug 05 18:56:15 http://lkml.iu.edu/hypermail/linux/kernel/1808.0/04276.html Aug 05 18:56:51 I think the ath10k changes have bin back ported. Aug 05 18:57:31 Or will be. Aug 05 19:11:55 it means to be better...? Aug 05 19:13:58 huaracheguarache: I'm using a tl-wdr3600 in client-mode (WDS/ 4addr) with option ieee80211w '2' in the wifi-iface stanza, that works Aug 05 19:15:43 karlp: https://i.imgur.com/MC7m8gD.jpg needs to run on the buildbots to be of more use though Aug 05 20:06:00 is it possible, that 309414ee8d6cf2e31476133606b2e390b0efbac5 (backported from master) contains a bug, as in it replaces $(LEDE_GIT) with $(PROJECT_GIT) which i can't find in lede-17.01 Aug 05 20:09:46 jow: ^ Aug 05 20:12:49 rotanid: the commit directly before that, bcf91e578c0fff183c34a52364ef44d78ee74d51 Aug 05 20:18:04 grml, my eyes... :D Aug 05 20:18:58 and did my grep in an old checkout on top of that. Aug 05 20:42:02 updated openwrt/upstream, https://sdwalker.github.io/uscan/index.html Aug 05 21:26:11 swalker: ok, and where's the scripts that made that though? can I even run it on my ownlogs? Aug 05 21:28:20 hi, on the topic of wpa3. Aug 05 21:28:30 I've seen that openwrt uses hostapd directly from git Aug 05 21:28:53 and more importantly, the used versions of hostapd seem to support wpa3 already Aug 05 21:29:08 however, openwrt's configuration system doesn't support wpa3 yet Aug 05 21:29:25 is someone working on adding wpa3 to the config system? Aug 05 21:44:36 Should changing the version of Perl in the Makefile cause all of the perl modules to get recompiled as well? Aug 05 21:59:59 est31: I don't think anyone is looking into that at the moment, patches would probably be very welcome Aug 05 22:00:47 pkgadd: I'm interested in sending patches Aug 05 22:01:15 pkgadd: thanks for the reply. Are anonymous contributions possible? Aug 05 22:02:09 est31: no, you'd have to sign your commits off, similar to linux kernel contributions (so real name and valid mail address), but you can submit the patches via mailing list or github pull requests Aug 05 22:02:38 https://openwrt.org/submitting-patches Aug 05 22:02:46 hmmm sorry that precludes me contributing then :/ Aug 05 22:05:27 karlp: https://pastebin.com/raw/DueXXU45 Aug 05 22:06:31 est31: while I understand the sentiment, it's not that painful ;) Aug 05 22:10:36 (I do suggest using a dedicated mail address for public contributions though) Aug 05 22:22:06 est31: why would you not want to put your name on your patches? Aug 05 22:23:14 I am not a dev so there mite be somthing I just don't know about! Aug 05 22:49:50 idk I just value a veil of anonymity Aug 05 22:51:31 people can always be deanonymized etc etc but I don't want to make it them THAT easy that I put my name next to every commit of mine Aug 05 22:53:07 ultimately I just want to scratch my own itch and use forward secrecy for my private home network. all devices on it are linuxes so as long as I figure out how to patch them all I can maybe achieve wpa3 only Aug 05 22:53:44 openwrt will get wpa3 support sooner or later, with or without me Aug 05 22:58:56 est31 I here you. Even if you did not want your name to be on your patches. I was just thinking the more people working on a progect I like the better. Aug 05 23:03:13 * fil is surprised that openwrt cannot deal with at least pseudonymous contributions -- I'm pretty sure Debian has a few contributors where only DAM (Debian Account Manager) has real ID info. Aug 05 23:04:57 if you look at recent commit logs, you will find several pseudonyms in the signoffs Aug 05 23:05:05 less than a handfull (of roundabout 1000 DDs) fall into that category (I only know of one such example), DAM does know their real name and it probably wasn't easy for them to achieve that status Aug 05 23:06:20 I am able to contribute to a multitude of open source projects without my clear name Aug 05 23:13:41 not the first time though that I've bumped into this roadblock Aug 05 23:13:53 * est31 shrugs, wishes everyone the best, and heads off Aug 05 23:53:12 theoretically one could always use a fictional real name ... Aug 05 23:53:43 Elvis has left the building Aug 05 23:53:49 i know **** ENDING LOGGING AT Mon Aug 06 03:00:00 2018